Toy models for varioius MCMC methods, standard Markov-chain Mote Carlo (MCMC), Parallel tempering (PT), and Hamiltonian Monre Calro (HMC). Each progaram draws samples in accordance with a certain probability distribution p(x,y) expressed as below:
- command:
bin/mcmc_toy
- output:
mcmc.out
- command:
mpirun -np 20 bin/pt_toy
- output:
pt.out
- command:
bin/hmc_toy
- output:
hmc.out
- Just type
make
- Needs
mpif90
that is linked togfortran
for PT
- Can be tweak in
src/params.f90
- Need re-compilation after changing parameters
- format: x, y, probability (w/o normalization constant)